Special Feature 603

Journal of Vegetation Science 10: 603-730, 1999© IAVS; Opulus Press Uppsala. Printed in Sweden

SPECIAL FEATURE

Plant functional types and disturbance dynamics

Editors

Sandra Lavorel & Wolfgang Cramer

Based on contributions presented at the GCTE-Workshop on Functional analysis of vegetation response to disturbanceheld in Montpellier, France, in March 1998, and, in part, the Session on Plant Functional Types and Plant Strategy Typesheld at the 41st Symposium of the International Association for Vegetation Science in Uppsala, Sweden, in July 1998.

ContentsMcIntyre, S., Díaz, S., Lavorel, S. & Cramer, W. — Plant functional types and disturbance dynamics –

Introduction

Philosophy and methodology for plant functional classifications

Weiher, E., van der Werf, A., Thompson, K., Roderick, M., Garnier, E. & Eriksson, O. — ChallengingTheophrastus: A common core list of plant traits for functional ecology

McIntyre, S., Lavorel, S., Landsberg, J. & Forbes, T.D.A. — Disturbance response in vegetation –towards a global perspective on functional traits

Pillar, V.D. — On the identification of optimal plant functional types

Results from field studies

Gitay, H., Noble, I.R. & Connell, J.H. — Deriving functional types for rain-forest treesDíaz, S., Cabido, M., Zak, M., Martínez Carretero, E. & Araníbar, J. — Plant functional traits, ecosystem

structure, and land-use history along a climatic gradient in central-western ArgentinaLavorel, S., McIntyre, S. & Grigulis, K. — Plant response to disturbance in a Mediterranean grassland:

How many functional groups?Hadar, L., Noy-Meir, I. & Perevolotsky, A. — The effect of shrub clearing and grazing on the

composition of a Mediterranean plant community: functional groups versus speciesLandsberg, J., Lavorel, S. & Stol, J. — Grazing response groups among understorey plants in arid

rangelandsKleyer, M. — Distribution of plant functional types along gradients of disturbance intensity and resource

supply in an agricultural landscapeDíaz Barradas, M.C., Zunzunegui, M., Tirado, R., Ain-Lhout, F. & Garcia Novo, F. — Plant functional

types and ecosystem function in Mediterranean shrubland

Using plant functional types to model ecosystem response to disturbance

Pausas, J.G. – Response of plant functional types to changes in the fire regime in Mediterraneanecosystems: A simulation approach

Campbell, B.D., Stafford Smith, D.M. & Ash, A.J. — A rule-based model for the functional analysis ofvegetation change in Australasian grasslands

604

609

621631

641

651

661

673

683

697

709

717

723

604 Special Feature

Abstract. Plant functional traits and types are useful conceptsin relation to disturbance responses of natural and managedecosystems. To explore their applicability in greater depth, aset of 12 papers presents a broad range of issues from method-ologies to the results of particular trait studies in the field, andmodelling approaches. So far, empirical studies have onlyallowed us to identify a few functional traits that are consist-ently associated with disturbance. To determine the trait vari-ations associated with climate, disturbance history and currentdisturbance regime as well as the interactions between thesefactors, global-scale comparisons of numerous individual stud-ies are required. Significant advances toward this ambitiousgoal are presented in these papers, and include: (1) the articu-lation of experimental and analytical methodologies for indi-vidual studies that could usefully contribute to a global com-parison; (2) the identification of core traits that can be used inthe further search for disturbance-related traits common to arange of environments; (3) further information on vegetationresponse to disturbance in terms of trait representation, and theidentification of attribute syndromes; (4) the identification ofissues for modelling disturbance dynamics using functionaltypes.

Keywords: Functional trait; Disturbance; Plant community.

Introduction

The search for plant functional types is the responseto a long-standing desire of ecologists to seek simplifiedpatterns in the richness of plants and the complexity ofecosystems (e.g. Raunkiær 1907). The search is beingcarried out for a range of purposes, e.g. for the system-atic analysis of ecosystem function, or for the improvedassessment of ecosystem sensitivity against changes inthe environment. However, the sets of traits or typesnecessarily differ among those applications (Woodward& Cramer 1996). Disturbance dynamics in natural andmanaged ecosystems have recently received increasedattention in empirical, as well as theoretical, studies. Ithas been recognized that diversity (or complexity) is

Plant functional types and disturbance dynamics – Introduction

McIntyre, S.1*, Díaz, Sandra2, Lavorel, Sandra3 & Cramer, Wolfgang4

1CSIRO Tropical Agriculture, 306 Carmody Rd, St Lucia, QLD 3067, Australia;2Instituto Multidisciplinario de Biología Vegetal, CONICET, Universidad Nacional de Córdoba, Facultad de

Ciencias Exactas, Físicas y Naturales, CC 495, 5000 Córdoba, Argentina;3Centre d’Ecologie Fonctionnelle et Evolutive, CNRS UPR 9056, 1919 Route de Mende,

F-34293 Montpellier Cedex 5, France;4Potsdam Institut für Klimafolgenforschung e.v. (PIK), Postfach 60 12 03, D-144 12 Potsdam, Germany;

*Corresponding author; Fax +61732142288; E-mail sue.mcintyre@tag.csiro.au

closely related to disturbance at several levels of scale.Our ability to address critical issues of diversity there-fore depends on our ability to understand disturbance.Investigating disturbance from a generic point of view(i.e. for more than one ecosystem) requires the identifi-cation of plant functional types that are directly relatedto the disturbance regime.

To advance research with respect to this goal, and tocome closer to a global synthesis, a workshop was heldin Montpellier, France, in March 1998, under the aus-pices of the International Geosphere Biosphere Pro-gramme (IGBP) and its core project Global Change andTerrestrial Ecosystems (GCTE). All key papers fromthis workshop, as well as related presentations from asession on Plant Functional Types and Plant StrategyTypes held during the 41st Symposium of the Interna-tional Association for Vegetation Science (IAVS) inUppsala, Sweden, in July 1998, are assembled in thisspecial issue.

While it is relatively easy to reach agreement on thegeneral aim of defining functional types for disturbanceassessments, the diversity of approaches that are possi-ble is bewildering, even within groups of scientists witha tightly defined philosophy. The research process in-volves a counterpoint between the creativity of indi-vidual efforts and the need for a unified, disciplinedapproach between individuals, over time. The twelvepapers presented in this issue represent this counterpointwell. The ideal of testing a unified set of traits (Westoby1998;Weiher et al. 1999) and using a uniform method-ology (McIntyre et al. 1999) is immediately challengedby the myriad of choices, restrictions and interestingavenues for discussion that faces each researcher in-volved in collection and presentation of data.

The authors in this issue present material that can bedescribed under the three topics with which the papers areprimarily concerned: methodologies (Pillar 1999; Gitayet al. 1999; McIntyre et al. 1999; Weiher et al. 1999),results of trait studies (Díaz et al. 1999; Díaz Barradas

Special Feature 605

et al. 1999; Hadar et al. 1999; Kleyer 1999; Landsberget al. 1999; Lavorel et al. 1999) and modelling (Campbellet al. 1999; Pausas 1999). Under these same headingswe discuss conclusions in common, issues raised andfinally, the implications of these writings for futuredirections.

Methodologies

Trait selection and measurement

Functional trait studies need to consider the choiceof traits to measure, how they are measured and how theresults are reported. Weiher et al. (1999) propose a listof core traits that could conceivably be tested over arange of environments, while McIntyre et al. (1999)outline how such a list could be developed in relation toa specific function (grazing response) including theneed to tailor a final trait list to the assemblage understudy. Both papers, and Díaz et al. (1999), advocate theuse of easily-measured traits (e.g. the structural traitsdescribed by Box 1996) over those that may be moreclosely linked to ecosystem function but are less likelyto be used, or to be measured in large numbers of species(e.g. chemical composition, decomposition rate, seedpersistence in the soil). It seems that a comparativeapproach to trait analysis will continue to depend oneasily-measured traits that are surrogates for function.However, the need to ultimately establish the linksbetween traits and function will also need to be ad-dressed. Díaz Barradas et al. (1999) measured traitsrelating to response to water stress and were able to linkthese with the distribution of species along a gradient ofwater table depth. A possible way forward would be tocalibrate easily-measured traits, which can be assessedin a high number of species, against more quantitativetraits, whose measurement is much more demanding,but whose ecosystem meaning is more direct or betterknown. Examples are the calibration of leaf toughnessagainst litter decomposition rate, or specific leaf area orleaf water content against relative growth rate.

Comparing functional analyses across different sitesand studies

McIntyre et al. (1999) argue that meaningful inter-pretations of functional traits will only result from thesynthesis of many studies. The use of common traitsacross studies as proposed by various authors (e.g. Hendry& Grime 1993; Westoby 1998; Weiher et al. 1999) willgo part way to achieving this. Even when the sametraits are reported in different studies, the measure-ment and definition of attributes becomes important to

the reader trying to compare studies. For example the‘medium’ and ‘large’ seed weight classes in Landsberget al. (1999) were both less than 1 mg and most specieswould have been classified as ‘light’ using the Lavorelet al. (1999) scheme. However, the fact that the at-tributes are defined in both papers makes this differenceapparent to the reader. For the same reasons it is impor-tant that authors report the entire profile of their studysystem, including the overall attribute make-up, thecomposition in terms major life forms (not all of whichmay have been analysed in detail) and traits that did notvary in the assemblage (McIntyre et al. 1999). Also,Díaz et al. (1999) stress that trait variation betweendifferent populations within the same species growingunder different grazing regimes may be substantial.This makes it difficult to extrapolate information onindividual taxa between regions, or even sites.

There are further problems for the reader trying tocompare and interpret published results from differentsites. Milchunas et al. (1988) eloquently argued for theimportance of considering disturbance history in theinterpretation of present-day disturbance response invegetation. More recently Balmford (1996) and Díaz etal. (1999) have reinforced this point. In most casesrelevant disturbance history can only be provided by theauthor and is not freely available to the reader. There-fore unless the disturbance history is appropriately de-scribed in published studies, the author forfeits thechance to contribute to the global picture of disturbance-related traits.

However, there are barriers to authors including theabove-mentioned information in publications. It is mostrelevant to inter-site comparisons, and not to the indi-vidual, immediate study being reported. Therefore thereis both a self-imposed and journal-imposed tendency toleave out basic descriptive information in manuscripts.This limits the potential use of published studies in latersyntheses which compare many published studies. Newways of exchange of scientific information, such as theelectronic networks established through programmessuch as IGBP-GCTE, provide alternative routes to morecomplete documentation, but they clearly also leavequestions open, such as the long time archive availabil-ity of electronically disseminated information.

Comparing different functional classifications of thesame plant assemblage

We have discussed the role of uniform methods andreporting in the comparison of data sets across differentsites. A less daunting, but still important task is to exam-ine the robustness of functional classifications relating toa particular plant assemblage. Formal approaches tothis have been described in Gitay & Noble (1997). In

606 Special Feature

this issue, Gitay et al. (1999) exemplify one of theseformal approaches by examining different classifica-tions of the same rainforest vegetation with respect togap dynamics. Congruency occurs when species in theassemblage are classified similarly using different charac-ter sets. The authors concluded that a morphologicalclassification, which was strongly related to phylogeny,was not relevant to dynamic behaviour. Instead, a deduc-tive classification based on an a priori theoretical view ofthe factors important in controlling gap dynamics wasmore successful.

Comparisons need to address not only classificationphilosophies and trait sets but also classification meth-ods per se. Pillar (1999) carried out a quantitativeassessment of the relative performance of different sub-sets of traits from a full trait list. A new approach basedon fuzzy set theory made it also possible to compareclassical classification with classifications introducingvarying degrees of fuzziness. Such optimization meth-ods open new avenues to address the spiny problem oftrait selection and to perform quantitative comparisonsacross classifications.

Less formal comparisons of a priori classificationsand response-related classifications are also reportedin Lavorel et al. (1999) and Hadar et al. (1999). Inthese studies, individual attributes were associated withdisturbance response. Using the same data describingvegetation response to disturbance, a priori functionaltypes were also tested for disturbance response. Again,the deductive approach was useful for data interpreta-tion and the identification of trait syndromes. Whentraits are analysed on an individual basis, the challengeis to identify combinations of significant attributes thatactually occur in the flora (syndromes). The testingand comparison of both individual attributes and apriori functional types assisted in the identification ofsyndromes.

Results of trait studies

Five papers in this issue present data in which re-sponse to disturbance is analysed in terms of plantattribute response. Although the general philosophiesbehind them were similar, they represent a diversity ofapproaches and traits sets over a range of systems. Traitsrelating to grazing response were examined in fourpapers (Díaz et al. 1999, Hadar et al. 1999, Landsberget al. 1999, Lavorel et al. 1999) with the focus being onthe herbaceous or understorey component of the vegeta-tion in most cases. A fifth paper addressed the identifi-cation of traits relevant to disturbance in agriculturallandscapes (Kleyer 1999).

These studies consistently identify the broad, al-

ready well-documented traits (e.g. prostrate growth formswith protected buds increase under heavy grazing), butfurther details and syndromes tend to differ across re-gions, even within a comparable climate. For example,in Israel the main grazing increaser groups in the herba-ceous vegetation of cleared shrublands were legumesand perennial geophytes (Hadar et al. 1999). On theother hand, in Portugal grasslands dominated by annu-als, receiving similar annual precipitation, small forbspecies with leafy stems were the only group favouredby grazing (Lavorel et al. 1999). Major differencesacross sites were interpreted as resulting from differen-tial availability of propagules at the landscape scale (alack of perennial propagules in the Portuguese land-scape), as well as from differences in grazing regimes(seasonal in Israel vs. continuous in Portugal).

Authors of the five papers agreed that plant func-tional classifications should search for syndromes (de-fined here as repeated attribute combinations that actu-ally occur in the flora), rather than identifying a list ofisolated traits. This is because a group of individualfunctional attributes identified in a data set may identify‘ideal’ functional combinations which may not actuallyexist in the vegetation. The description of syndromeshas been possible in most cases (Díaz et al. 1999; Hadaret al. 1999; Lavorel et al. 1999). The exception wasLandsberg et al. (1999) who identified a number oftraits negatively or positively associated with grazingpressure, but failed to find strong recurrent patterns ofassociation among them that would match associationsin the local flora. An alternative approach is followed byKleyer (1999) who first identified syndromes repre-sented in the flora and then assigned functionality tothese syndromes through an analysis of their distribu-tions along disturbance and productivity gradients.

Modelling

At some level, functional type classifications will beincorporated into theoretical frameworks that allow theuse of a predictive mode, in other words, numericalsimulation models (Cramer 1997). Two papers describevery different modelling approaches which demonstratedapplications for functional classifications based on dis-turbance and climate response. Pausas (1999) used fourfunctional types based on regeneration and growth at-tributes to predict response to fire in Mediterraneanshrublands using two simulation models. To the extentthat the predictions could be verified (which is limited),the modelled results were accurate, and consistent be-tween models. Campbell et al. (1999) described theconceptual stages of model development for a widerrange of vegetation types, climates and land uses.

Special Feature 607

In general, moderately simple classifications withfew attributes seem desirable for modelling applica-tions. Simple models based on these seem to havemoderate success, supporting the premise that func-tional types are useful for modelling. This at least iscompatible with progress in functional type identifica-tion so far. At a global scale, it has been possible toidentify only very few traits that are consistently associ-ated with disturbance (Lavorel et al. 1997). Campbell etal. (1999) concur with Lavorel et al. (1997) on the needfor functional types to be tailored to a specific purposeor function e.g. response to climate or response to dis-turbance. There is also a need for further developmentsin process-based modelling of plant functional types indifferent environments.

Several issues were found to be in common withboth field-based studies and modelling. First, is the needto further understand which attributes are linked withspecific climate and disturbance regimes. Second is thequestion of how to relate a set of individual functionalattributes to actual vegetation (i.e. what syndromes arefunctionally significant and therefore what are the plantfunctional types). This problem is intensified by the lackof congruency between regeneration traits and traits ofthe established phase within the same plant assemblage(e.g. Leishman & Westoby 1992; Díaz & Cabido 1997).Both types of traits are useful for modelling applications.Finally, modellers need assembly rules to describe thecoexistence of different plant functional types in vegeta-tion, a problem which field-based researchers are tack-ling from the bottom up. Field work provides essentialinformation to formulate ‘sensible’ assembly rules thatare constrained by what actually happens in nature.

Some unanswered questions

The fact that most findings connecting plant traitsand specific disturbances so far appear to be ‘trivial’ or‘expected’ does not invalidate the search for functionaltypes relating to climate or disturbance response. Thereis still a need for formal comparison of many studies ata global scale, something which still appears to belacking. Also, climate and disturbance history are fac-tors that operate at higher organizational levels in bioticsystems (sensu Allen & Starr 1982) and interact withpresent-day disturbance regimes in complex ways. Thevast majority of the literature linking disturbance withplant traits to date involve individual case studies. Inthose cases, climate and disturbance history are as-sumed as fixed, and therefore hardly addressed. In glo-bal cross-comparisons, however, these two sources ofvariation have to be formally incorporated.

The following list of questions are some issues that

are relevant to the global cross-comparison of trait stud-ies. To propose formal methods for the comparison ofclassifications across sites is beyond the scope of thissummary. However, the following questions may serveto sharpen thinking and assist the development of such acomparison:

1. Which traits are consistently associated with particu-lar disturbances across a wide range of climates anddisturbance histories? Only a small number of traitshave been identified to date, but have enough traits beentested at enough sites to accept this result? The coretraits identified by Hendry & Grime (1993), Westoby(1998) and Weiher et al. (1999) provide a starting pointfor this question. These are traits identified as likely tobe linked to vegetation dynamics in a general sense.Those that are relevant to most vegetation, easy tomeasure, and that have well-articulated methodologiesproducing consistent measurements, are most likely tobe widely tested in a range of environments.

2. How do climate or disturbance histories affect veg-etation composition in terms of major life forms? Thelong-terms effects of climate and disturbance regime areprimary determinants (‘filters’, cf. Díaz et al. 1999) ofvegetation composition, and account for major differ-ences in form and structure. Major life forms (e.g. trees,shrubs, grasses, annual forbs) are a simple way to ac-count for variation, and identifying vegetation that iscomparable at finer levels of detail e.g. there may be fewrelevant traits in common between tree-dominated veg-etation and annual grasslands.

3. How does vegetation of comparable life forms, cli-mate and disturbance history respond to current distur-bance? When comparable data sets are identified, amore detailed analysis of trait response data will bepossible. Responses to different disturbance types shouldbe considered separately. A sub-question, touched uponby Pillar (1999), is: What is the minimum number oftraits we need to predict responses to disturbance his-tory/ climate/ current disturbance combinations?

4. Does climate act more strongly on physiologicaltraits and disturbance history act more strongly onregeneration traits? The apparent lack of correlationbetween vegetative and regeneration traits may relate tothe long-term selection pressures of climate and distur-bance, acting separately.

Attempting to address these and related questionsmay assist our progress towards finding some generalisa-tions linking plant traits and disturbance at a worldwidescale. The challenge here is in providing knowledgethat represents a substantial advance with respect to tradi-

608 Special Feature

tional and well known life-form classifications. This mustbe balanced with efforts to keep the whole approach assimple, parsimonious, and realistic as possible.

References

Allen, T.F.H. & Starr, T.B. 1982. Hierarchy: perspectives forecological stability. University of Chicago Press, Chi-cago, IL.

Balmford, A. 1996. Extinction filters and current resilience:the significance of past selection pressures for conserva-tion biology. Trends Ecol. Evol. 11: 193-196.

Box, E.O. 1996. Plant functional types and climate at theglobal scale. J. Veg. Sci. 7: 309-320.

Campbell, B.D., Stafford Smith, D.M. & Ash, A.J. 1999. Arule-based model for the functional analysis of vegetationchange in Australasian grasslands. J. Veg. Sci. 10:723-730. (This issue.)

Cramer, W. 1997. Using plant functional types in a globalvegetation model. In: Smith, T.M., Shugart, H.H. & Wood-ward, F.I. (eds.) Plant functional types: their relevance toecosystem properties and global change, pp 271-288.Cambridge University Press, Cambridge.

Díaz, S. & Cabido, M. 1997. Plant functional types andecosystem function in relation to global change: a multiscaleapproach. J. Veg. Sci. 8: 463-474.

Díaz, S., Cabido, M., Zak, M., Martínez Carretero, E. & Araníbar,J. 1999. Plant functional traits, ecosystem structure, andland-use history along a climatic gradient in central-westernArgentina. J. Veg. Sci. 10: 651-660. (This issue.)

Díaz Barradas, M.C., Zunzunegui, M., Tirado, R., Ain-Lhout,F. & Garcia Novo, F. 1999. Plant functional types andecosystem function in a Mediterranean shrubland. J. Veg.Sci. 10: 709-716. (This issue.)

Gitay, H. & Noble, I.R. 1997. What are functional types andhow should we seek them? In: Smith, T.M., Shugart, H.H.& Woodward, F.I. (eds.) Plant functional types: theirrelevance to ecosystem properties, pp. 3-19. CambridgeUniversity Press, Cambridge.

Gitay, H., Noble, I.R. & Connell, J.H. 1999. Deriving func-tional types for rain-forest trees. J. Veg. Sci. 10: 641-650.(This issue.)

Hadar, L., Noy-Meir, I. & Perevolotsky, A. 1999. The effect ofshrub clearing and grazing on the composition of a Medi-

terranean plant community: functional groups versus spe-cies. J. Veg. Sci. 10: 673-682. (This issue.)

Hendry, G.A.F. & Grime, J.P. 1993. Methods in comparativeplant ecology: a laboratory manual. Chapman & Hall,London.

Kleyer, M. 1999. Distribution of plant functional types alonggradients of disturbance intensity and resource supply in anagricultural landscape. J. Veg. Sci. 10: 797-708. (This is-sue.)

Landsberg, J., Lavorel, S. & Stol, J. 1999. Grazing responsegroups among understorey plants in arid rangelands. J. Veg.Sci. 10: 683-696. (This issue.)

Lavorel, S., McIntyre, S. & Grigulis, K. 1999. Plant responseto disturbance in a Mediterranean grassland: How manyfunctional groups? J. Veg. Sci. 10: 661-672. (This issue.)

Lavorel, S., McIntyre, S., Landsberg, J. & Forbes, T.D.A.1997. Plant functional classification: from general groupsto specific groups based on response to disturbance. TrendsEcol. Evol. 12: 474-478.

Leishman, M.R. & Westoby, M. 1992. Classifying plants intogroups on the basis of associations of individual traits-evidence from Australian semi-arid woodlands. J. Ecol.80: 417-424.

McIntyre, S., Lavorel, S, Landsberg, J. & Forbes, T.D.A.1999. Disturbance response in vegetation – towards aglobal perspective on functional traits. J. Veg. Sci. 10:621-630. (This issue.)

Milchunas, D.G., Sala, O.E. & Lauenroth, W.K. 1988. Ageneralized model of the effects of grazing by large her-bivores on grassland community structure. Am. Nat. 132:87-106.

Pausas, J.G. 1999. Response of plant functional types tochanges in the fire regime in Mediterranean ecosystems: Asimulation approach. J. Veg. Sci. 10: 717-722. (This issue.)

Pillar, V.D. 1999. On the identification of optimal plant func-tional types. J. Veg. Sci. 10: 631-640. (This issue.)

Raunkiær, C. 1907. Planterigets Livsformer. GyldendalskeBoghandel & Nordisk Forlag, København / Kristiania.

Weiher, E., van der Werf, A., Thompson, K., Roderick, M.,Garnier, E. & Eriksson, O. 1999. Challenging Theophrastus:A common core list of plant traits for functional ecology.J. Veg. Sci. 10: 609-620. (This issue.)

Westoby, M. 1998. A leaf-height-seed (LHS) plant ecologystrategy scheme. Plant Soil 199: 213-227.

Woodward, F.I. & Cramer, W. 1996 Plant functional types andclimatic changes: Introduction. J. Veg. Sci. 7: 306-308.